The requirement for an implementation is understandable but is a departure from MPI 1 and MPI 2 - there we only required a convincing argument that the requirements were possible, not an example code. I think this worked reasonably well (with some pluses and minus) and believe that the MPI 1 and MPI 2 rule should be sufficient. Bill On Dec 7, 2010, at 7:29 PM, Graham, Richard L. wrote:

The attached document is a summary of our discussion yesterday - we will talk about this tomorrow (Wed) as see if there are updates that should be made. Please come prepared with comments.

Thanks, Rich <standardization_process.docx><ATT00001..txt>

William Gropp Deputy Director for Research Institute for Advanced Computing Applications and Technologies Paul and Cynthia Saylor Professor of Computer Science University of Illinois Urbana-Champaign